An Out-of-Sample Evaluation of Dynamic Portfolio Strategies

Abstract

This article evaluates out-of-sample portfolio performance for a real-time investor who can exploit time variation in the conditional mean and volatility of stock returns in optimizing a multiperiod portfolio choice problem. With the presence of parameter uncertainty, our out-of-sample analysis shows that ignoring time variation in the first two return moments leads to significant utility costs of at least 1.97% of annualized certainty equivalent return. Accounting for the time-varying risk premium plays a more important role than considering time-varying volatility in improving portfolio performance. Interestingly, behaving myopically or ignoring the hedge against changes in future investment opportunities can lead to small out-of-sample utility losses or even utility gains.